Partial Least Squares (PLS) is a regression techniques used in cases where there is high predictor correlation.
Partial Least Squares (PLS) is a regression technique used in cases where there is high predictor correlation. PLS works similar to PCA in that it finds linear combinations of predictors called latent variables. The underlying algorithms maximizes the summary of both the covariance and the response, thereby capturing variation and predictive relationships between the predictors and the response. Similar to PCA, PLS has a tuning parameter for the number of components (latent variables) to retain.
